TY - JOUR
T1 - Mining and indexing graphs for supergraph search
AU - Yuan, Dayu
AU - Mitra, Prasenjit
AU - Giles, C. Lee
N1 - Copyright:
Copyright 2017 Elsevier B.V., All rights reserved.
PY - 2013/8
Y1 - 2013/8
N2 - We study supergraph search (SPS), that is, given a query graph qand a graph database G that contains a collection of graphs, returngraphs that have q as a supergraph from G. SPS has broad applicationsin bioinformatics, cheminformatics and other scientific andcommercial fields. Determining whether a graph is a subgraph (orsupergraph) of another is an NP-complete problem. Hence, it is intractableto compute SPS for large graph databases. Two separateindexing methods, a "filter + verify"-based method and a "prefixsharing"-based method, have been studied to efficiently computeSPS. To implement the above two methods, subgraph patterns aremined from the graph database to build an index. Those subgraphsare mined to optimize either the filtering gain or the prefix-sharinggain. However, no single subgraph-mining algorithm considersboth gains. This work is the first one to mine subgraphs to optimize boththe filtering gain and the prefix-sharing gain while processing SPSqueries. First, we show that the subgraph-mining problem is NPhard. Then, we propose two polynomial-time algorithms to solvethe problem with an approximation ratio of 1-1/e and 1/4 respectively. In addition, we construct a lattice-like index, LW-index, toorganize the selected subgraph patterns for fast index-lookup. Ourexperiments show that our approach improves the query processingtime for SPS queries by a factor of 3 to 10.
AB - We study supergraph search (SPS), that is, given a query graph qand a graph database G that contains a collection of graphs, returngraphs that have q as a supergraph from G. SPS has broad applicationsin bioinformatics, cheminformatics and other scientific andcommercial fields. Determining whether a graph is a subgraph (orsupergraph) of another is an NP-complete problem. Hence, it is intractableto compute SPS for large graph databases. Two separateindexing methods, a "filter + verify"-based method and a "prefixsharing"-based method, have been studied to efficiently computeSPS. To implement the above two methods, subgraph patterns aremined from the graph database to build an index. Those subgraphsare mined to optimize either the filtering gain or the prefix-sharinggain. However, no single subgraph-mining algorithm considersboth gains. This work is the first one to mine subgraphs to optimize boththe filtering gain and the prefix-sharing gain while processing SPSqueries. First, we show that the subgraph-mining problem is NPhard. Then, we propose two polynomial-time algorithms to solvethe problem with an approximation ratio of 1-1/e and 1/4 respectively. In addition, we construct a lattice-like index, LW-index, toorganize the selected subgraph patterns for fast index-lookup. Ourexperiments show that our approach improves the query processingtime for SPS queries by a factor of 3 to 10.
UR - http://www.scopus.com/inward/record.url?scp=84891094619&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84891094619&partnerID=8YFLogxK
U2 - 10.14778/2536206.2536211
DO - 10.14778/2536206.2536211
M3 - Article
AN - SCOPUS:84891094619
VL - 6
SP - 829
EP - 840
JO - Proceedings of the VLDB Endowment
JF - Proceedings of the VLDB Endowment
SN - 2150-8097
IS - 10
ER -